In 1982, AT&T was split up by the American government. In Japan, the state is now considering a similar policy towards NTT. In both cases, the companies have strenuously resisted a massive government intervention pursued in the name of deregulation. But in the future, divestiture may not be state-imposed at all but rather company-initiated, and it may be in the telecommunications companies' own interest to split themselves up. AT&T just did so a few days ago in a second and voluntary divestiture.

All this is part of the logic of transformation in telecommunications, in which service competition leads to infrastructure competition, which in turn leads to interconnection, unbundling, systems integration, and eventually radical corporate restructuring.

With liberalization of entry, multiple networks emerge. They must become linked with each other through various interconnection arrangements. Interconnection, in turn, is fairly meaningless without multiple physical interfaces. If an incumbent network offers an entrant interconnection at a far-off point, little is resolved. Thus, interconnection and unbundling of network functionalities into modules go hand-in-hand.

How can the numerous network hardware and software modules be integrated into a usable whole? Perhaps the most promising approach is systems integration, both by independent firms and by carriers. The characteristics of pure systems integraters is that they do not own or operate the various sub-production activities but rather select optimal elements in terms of price and performance, package them together, manage the bundles, and offer it to the customer on a one-stop basis.

These systems access into each other, leading the telecommunications environment to evolve from the "network of networks," in which carriers interconnect, to the "system of systems," in which systems integrators link up. The Internet is a step in that direction.

Interconnection leads to unbundling and systems integration, and this in turn leads carriers to reorganize themselves, based primarily on the business logic of letting a business segment seek its own suppliers independently of its other corporate affiliates if it wants to survive in a highly competitive market. When there are seperate markets for seperate modules, it is every module financially on its own bottom, in the long run. Each module provider must buy, sell, or joint-venture with modules competing with its own parent company, if the rival offers better terms. This creates major centrifugal forces inside the organization which in time will make coherent strategies difficult.

This must inevitably lead to the need for reorganizing the firms providing the modules. There are two main avenues. First, separating the firm into a set of independent business units for different modules or module-combinations. And second, breaking up the firm separating modules from the systems integrators, each other, and from part of the business. The strategy would depend on different factors: the synergies (economies of scope) vs. the diseconomies of scope; the extent of regulation and restrictions on various modules; different anti-monopoly and anti-concentration rules; organizational cultures; and turf battles inside the firm.

This is not mere hypothesis. In the US, Pacific Telesis reorganized itself in 1994 in a major voluntary self-divestiture, spinning off its mobile subsidiary that receives no "fraternal" preferences. The Rochester Telephone Co. separated itself into a network company offering transmission to all, including service competitors, as well as a services operator offering the actual service to customers. And now, AT&T, once split by government mandate into eight pieces, is separating itself voluntarily into four independent parts. Its chairman, Robert Allen, argues that "the complexity of trying to manage these different businesses began to overwhelm the advantages of our integration... Conflicts have arisen, and each of our businesses has to react more quickly." Traditional PTOs may try to delude themselves that AT&T's second divestiture is about computers and equipment, not networks. But that is a distinction without a big difference. The economic point is that part of the company is harmed by another part competing against its best customers. The same dynamics will affect different network modules in a competitive environment. The widening of national markets inside the European Union will delay but not contain these pressures.

At present, many media firms are engaging in high-priced empire building invoicing vague synergies. In some cases, the theory seems to be that the mating of two elephants results in a tiger rather than a dinosaur. But in a competative future, company attention will have to shift to a radical restructuring and focusing. Liberalization is just the beginning.